As an important carrier of national culture, national patterns have rich historical information and unique aesthetic value, but in modern design, traditional design methods are difficult to meet the innovative and personalized needs. In this study, particle swarm optimization (PSO) algorithm is introduced into pattern design optimization. The design parameters are transformed into computable vectors by particle coding, and the pattern quality is evaluated by fitness function, and the innovation and aesthetic improvement of the pattern are realized by optimizing the particle position. At the same time, combined with collaborative filtering technology, according to users’ historical behavior and preferences, personalized pattern design schemes are recommended for users to meet users’ needs for personalized design. The experimental results show that the PSO optimization group and the PSO combined with collaborative filtering group are superior to the traditional design group in innovation, aesthetics and style consistency. Among them, the combination of PSO and collaborative filtering has the highest score in all evaluation indexes, which shows that this method can significantly improve the innovation and aesthetics of design, and at the same time enhance the consistency of style. The survey of users’ satisfaction shows that the system performs well in terms of design efficiency, personalized recommendation accuracy and system ease of use, with an average score of 4.2, 4.5 and 4.0 respectively, and high satisfaction accounts for 82%, 90% and 75% respectively. PSO algorithm achieves fitness convergence within 100 iterations, the real-time response time of collaborative filtering recommendation is less than 0.5 s, and the accuracy (Top-5 recommendation hit rate) reaches 78%.